parallel-agents
Pass
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: SAFENO_CODE
Full Analysis
- SAFE (SAFE): No malicious patterns, obfuscation, or high-risk command execution were found within the provided skill file.
- NO_CODE (SAFE): The skill consists exclusively of documentation and orchestration metadata, with no included scripts or executable code.
- Indirect Prompt Injection (SAFE): The skill describes patterns for passing context between agents (e.g., passing findings from a 'security-auditor' to a 'test-engineer'). While this constitutes an attack surface for indirect prompt injection, the risk is mitigated by the restricted toolset (Read, Glob, Grep) which prevents destructive actions. Evidence chain: 1. Ingestion point: Synthesis and context passing between specialized agents. 2. Boundary markers: Absent in provided patterns. 3. Capabilities: Limited to read-only operations. 4. Sanitization: Not explicitly implemented in the skill.
Audit Metadata